GRAMMATICALIZATION

1. Process of grammaticalization proceeds from discourse down to sentence or morphosyntactic level. (Hopper 1979, 1982). Hopper also deals with aspect marking, arguing that completion (assoc. with perfective aspect) derives from a "more fundamental discourse function" i.e. signallying successive events in narration.

   Herring's article focuses on this claim, and makes following assertions.

   • viDu has been described as expressing completion, and is most difficult to characterize: completive, definitive, intensive.

   • Claims it has a different meaning when in 'conjunctive' (AVP) form, signalling 'disjunction' and separating clauses into discrete, unrelated events.

   • Says this 'recalls' the lexical vb viDu 'leave, leave off.'

   • Quotes Annamalai (1985) who identifies another auxiliary viDu

     extending "transitive meaning 'let' to that of 'release (of direct object following the action of the main verb)', and by extension, 'release into the hands of an intermediary."

     These two are different phonologically, the latter not showing deletion of initial [v] the way the aspectual one does. [So why doesn't SH show us this deletion in her examples? hs]

   • Aspectual viDu is the 'more grammaticalized of the two' and is the one SH is dealing with.

2. SH then proceeds to try to characterize aspectual viDu semantically:

   • Cites the usual descriptions:

     • Completive (in S's in isolation)

     • Definiteness or assurance

     • Emphasis on the proposition expressed by main vb

     • Unexpectedness or inadvertence;

     • Expected, long-awaited outcome

     • interacts differentially with tense, so for some analysts, it's completive in the past and definitive in the non-past.

     • NB: SH doesn't mention problems with negation and differential occurrence with imperative S's. (hs)

   • Evidence for completive: Tamil past tense-marked transitive-accomplishment verbs imply change of state "only weakly" so that indication that 'X did Y' followed by 'But Y didn't happen' is not contradictory.

     HS: my own analysis of this is that completion of change-of-state is not assumed and must be stated explicitly (if there is doubt), especially in exdeictic situations where action occurs elsewhere, off-stage, and results are not known or knowable to Ego. For verbs like vaa 'come' this is not a problem, so use of (v)iDu is rarer, or has more of the 'definitive' meaning.

   • SH says 'completive' is fine for 'transitive-accomplishment' verbs, but not so for intransitive vbs and 'verbs of achievement, process, and state'; we can't say process of 'going home' is more completed than others.

     HS: again, her S. 11, 'Kumar went home' is precisely the example of an exdeictic situation (movement away from Ego) where we don't know the outcome; I find such sentences 'unnatural' or incredibly vague, and very uncommon in spoken Tamil. (Meaning could be construed as 'Kumar started out to go home but we don't know whether he actually arrived' and S 11.b does give more information, e.g. 'Kumar definitely got home'.

   • SH also claims that if you really want to show completion, you'll use muDi 'finish' or muDicc-iDu or by use of adverbials.

3. Definitive analysis:

   • Most analysts say non-past only;

   • But most non-past 'meanings' are pragmatic, not semantic (a promise in 14, a warning in 15).

   • plus, viDu not incompatible with elements that express uncertainty or lack of definiteness, as in (17).

     HS: yes, but one could also gloss 17 as "it is a definite possibility that Kumar will perhaps come;" or "Kumar could definitely make an appearance/show up." SH doesn't gloss vandiDulaam but it is a possible sentence; if it doesn't mean what she says, what does it mean?

   Having thus demolished (or weakened) the two principal semantic interpretations of viDu she goes on to her own claim, which is that its meaning is mainly pragmatic and driven by discourse. This is of course a possibility, but why must it be the only interpretation? SH also fails to give examples of the problems with

   • negative sentences (where viDu is rare)
   • imperative contexts (incl. negative imperatives) where viDu is common;
   • other issues we have addressed.

4. Her next section (2.2, viDu in discourse) proposes to provide extensive contextualization;

   • First SH repeats the various characterizations, shown above

   • Wants to show that viDu has a perfect(ive) meaning, but ignores/dismisses the role of iru in 'perfect' constructs.

   • Cites Hopper to show how perfective aspect plays a role in complex of

     "of notions cross-linguistically, including strict chronological sequence of dynamic, kinetic events; human topics, preservation of [presupposed?] subject, ... assertion of new info in the vb; and foregrounding or signalling of events indispensible to the narrative."

   • SH then tests this with an analysis of a 20 narratives by 9 speakers. Finds that 20% of verbs are modified by viDu , of which most are intransitive accomplishments verbs inflected for past tense. Conclusion: Tamil viDu is "strong indicator" of narrative perfectivity (as characterized by Hopper) and that " the evidence supports the claim that the primary function of viDu in narrative discourse is that of perfectivity."

   • (HS: yeah, but is all Tamil narrative discourse?)

5. Another assertion: Perfectivity as Core Function Later she says that since Tamil has a productive past tense which implies completion (HS: earlier she shows that it does this so weakly as to be contradict this assertion), the interpretation of completive is "somewhat restricted in the language" (p. 288-9).

   Here SH seems to be claiming two things:

   • Importance of viDu in narrative discourse (HS: no problem)
   • Lack of other functions (or at least minimal importance of other functions previously attributed to it (HS: Big Problem! >
   • Narrative discourse as the origin (sic!) for all the major categories of meaning expressed by auxiliary viDu (including the ones she seems to be saying don't really exist) on the propositional as well as the discourse level.(p. 289-90)

6. Conclusion: her claims are too strong.